EDITORIAL COMMENTARY Detection of drug-induced proarrhythmia: Balancing preclinical and clinical studies Marek Malik, PhD, MD From the Department of Cardiac and Vascular Sciences, St. George’s Hospital Medical School, London, United Kingdom. Assessment of proarrhythmic toxicity and the likelihood of torsades de pointes (TdP) induction presently is required as an integral part of the development of practically every pharmaceutical agent. 1 Despite substantial advances, 2 the tools currently available for such an assessment are neither highly precise nor entirely understood. Consequently, the perception of the requirement to test for proarrhythmia occasionally involves skepticism and distrust. Are there “good” and “bad” drugs? In terms of proarrhythmia and TdP induction, it clearly is not useful to classify drugs simply as good and bad. Every case of drug-induced TdP is a combination of drug action and the patient’s susceptibility. Different drugs can be por- trayed only on a continuous scale of “how much” the pa- tient’s susceptibility is needed. On one side of the spectrum are antiarrhythmic drugs that cause TdP fairly frequently (e.g., quinidine with a reported TdP incidence of 5% of all treated patients 3,4 ). On the other side of the spectrum are drugs that are generally very safe but may cause TdP in peculiarly susceptible patients (e.g., fexofenadine, which, despite being a safe drug generally, caused TdP reproduc- ibly in a particular published case 5 ). This continuous scale of proarrhythmic toxicity “levels” likely is linked not only to the multitude of ion channels responsible for cardiac electrophysiology but also to their distribution within the myocardium and to the geometry of distribution of cells with different electrophysiologic prop- erties. Although these distributions likely are congenital, they are not necessarily genetically determined but might, similar to papillary lines of a fingerprint, depend on pure embryonic chance. Huge numbers of combinations of elec- trophysiologic distributions exist, and our present under- standing of congenital long QT and Brugada syndromes might only be the tip of a big iceberg. Without much exaggeration, perhaps one can claim that for every chemical more complex than water or sodium chloride, there is a patient in whom the chemical might trigger arrhythmia, especially if it is combined with other factors, such as electrolyte disturbance or subclinical ischemia of a peculiar myocardial region. At this stage of understanding, it seems unreasonable to aim at characterizing the proarrhythmic “levels” of drugs that cause arrhythmia at any frequency 0, that is, includ- ing solitary and highly unusual cases. Only in the future (and probably not very immediate future) might it be pos- sible to characterize the proarrhythmic levels of an “indi- vidual drug–individual patient” combination. At present, a threshold of “regulatory awareness” must exist, that is, a selection of proarrhythmic level must be made such that drugs that cause proarrhythmia less frequently are consid- ered “safe” and cases in which TdP eventually occurs while the patient is taking these drugs are “blamed” on patient susceptibility rather than on drug action. Experience with regulatory attitude concerning drugs such as terfenadine, 6 cisapride, 7,8 or moxifloxacin 9,10 sug- gests that the present threshold of regulatory awareness is approximately one case of proarrhythmia per approximately 10 5 –10 6 clinical exposures. The present regulatory approach requires careful risk-to-benefit analysis of all drugs that have higher levels of proarrhythmic risk, whereas drugs that have risk below the threshold are considered sufficiently safe so that their proarrhythmic potential can be ignored, at least for regulatory and drug-labeling purposes. Given the present understanding of the problem, this approach is fully reasonable. Regulatory reflection of tests for drug-induced proarrhythmia Because no clear distinction between “good” and “bad” drugs is possible, regulatory decisions must be based on some arbitrary limits. Although this approach is understand- able and appropriate, the existence of consensus-driven thresholds is not scientifically helpful for the development of simple tests that would closely approximate regulatory requirements. Perhaps this is one reason why, despite useful standardization initiatives, some discord exists between the Address reprint requests and correspondence: Dr. Marek Malik, Department of Cardiac and Vascular Sciences, St. George’s Hospital Med- ical School, London SW17 0RE, England. E-mail address: m.malik@sghms.ac.uk. 1547-5271/$ -see front matter © 2005 Heart Rhythm Society. All rights reserved. doi:10.1016/j.hrthm.2005.04.018